Larval food limitation in butterflies: effects on adult resource allocation and fitness
- First Online:
- Cite this article as:
- Boggs, C.L. & Freeman, K.D. Oecologia (2005) 144: 353. doi:10.1007/s00442-005-0076-6
- 708 Views
Allocation of larval food resources affects adult morphology and fitness in holometabolous insects. Here we explore the effects on adult morphology and female fitness of larval semi-starvation in the butterfly Speyeria mormonia. Using a split-brood design, food intake was reduced by approximately half during the last half of the last larval instar. Body mass and forewing length of resulting adults were smaller than those of control animals. Feeding treatment significantly altered the allometric relationship between mass and wing length for females but not males, such that body mass increased more steeply with wing length in stressed insects as compared to control insects. This may result in changes in female flight performance and cost. With regard to adult life history traits, male feeding treatment or mating number had no effect on female fecundity or survival, in agreement with expectations for this species. Potential fecundity decreased with decreasing body mass and relative fat content, but there was no independent effect of larval feeding treatment. Realized fecundity decreased with decreasing adult survival, and was not affected by body mass or larval feeding treatment. Adult survival was lower in insects subjected to larval semi-starvation, with no effect of body mass. In contrast, previous laboratory studies on adult nectar restriction showed that adult survival was not affected by such stress, whereas fecundity was reduced in direct 11 proportion to the reduction of adult food. We thus see a direct impact of larval dietary restriction on survival, whereas fecundity is affected by adult dietary restriction, a pattern reminiscent of a survival/reproduction trade-off, but across a developmental boundary. The data, in combination with previous work, thus provide a picture of the intra-specific response of a suite of traits to ecological stress.
In holometabolous insects, adult morphology and fitness are influenced by the availability and allocation of resources from larval and adult feeding. Larval feeding first supports growth of the larval insect. At some point, growth stops, and accumulated larval nutrients are allocated during metamorphosis into the adult body plan. The resulting adult individual has a given body size and given allometric relations among various body parts, such as wings, flight muscles, ovaries, and so on. Additionally, some nutrients may be allocated to “storage‘’, in the form of lipid or storage proteins for example. The adult insect itself may feed (although not all species do). The combination of incoming adult nutrients (“income”) and stored larval nutrients (“capital reserves”) is then used in specific proportions to support the adult functions of reproduction, survival, and dispersal.
Variation in the quality or quantity of larval or adult food can result in alterations to the normal pattern of allocation. We can thus vary the larval or adult diets to gain insight into the organization of allocation, by examining the resulting changes in life history and/or morphological traits. Some changes in quality or quantity of food may have severe impacts on life history traits, particularly if larval and adult diets provide different nutrients (e.g., Boggs and Ross 1993; O’Brien et al. 2005). In contrast, other changes in quality or quantity of food may have little or no effect. This would occur, for example, if temporary decreases in adult food availability are buffered by increased usage of larval reserves (with later compensatory feeding by adults), or if nutrients provided to the female by males at mating can compensate for decreased availability of female reproductive stores from larval feeding. Thus, understanding the effects of resource variation can aid in predicting the extent to which a species’ life history traits are buffered against environmental variation, with consequent buffering of fitness and population dynamics (Boggs 2003).
The allocation fate of larval resources during metamorphosis is particularly critical to the insect’s fitness, since it both determines adult morphology and contributes to adult life history traits, and thence fitness as outlined above. Such allocation results in integrated suites of specific adult morphological, life history, and foraging traits. These have been documented among species for Lepidoptera and Hymenoptera (Boggs 1981; Boggs 1986; Karlsson and Wickman 1989; Rivero et al. 2001). Increased allocation of larval resources to body (as opposed to reproduction) is associated with qualitatively and quantitatively richer adult diets, increased survival, an increasing reliance on adult food for reproduction and a more drawn out age-specific fecundity pattern.
However, such adult trait suites reflecting trade-offs associated with larval resource allocation have not been studied intra-specifically to the same extent. Our understanding is more or less limited to the observation that a reduction in larval food quality or quantity generally results in reduced adult body size and fat content with consequent negative effects on fecundity (e.g., Scriber and Slansky 1981; Fischer and Fiedler 2001; Ernsting et al. 1992; Awmack and Leather 2002), or on male reproductive success (e.g., Carroll 1994; Delisle and Hardy 1997). While extremely useful, this information does not yet provide insight into the organization of allocation as a whole within any one holometabolous organism, or the integrated suites of adult life history and morphological traits that result.
Here we document the effects of larval food restriction in a butterfly, Speyeria mormonia Edwards (Lepidoptera: Nymphalidae) on resulting adult morphology and fitness parameters. As context, we already have considerable information concerning the effects of adult food restriction on reproduction and survival in this species. Reproduction declines linearly with reductions in adult nectar intake, with oocytes resorbed and their nutrients re-allocated (Boggs and Ross 1993). Adult survival is unaffected by adult nectar deprivation (down to 33% of ad libitum feeding). The observed effect on reproduction is reinforced by radiotracer and stable isotope studies showing that 80% of egg carbon is derived from adult nectar feeding (Boggs 1997; O’Brien et al. 2004), even including synthesis of egg non-essential amino acids from nectar carbon (O’Brien et al. 2005). Most of the larval carbon in eggs is in the form of essential amino acids (O’Brien et al. 2002; O’Brien et al. 2004, 2005).
Given the above information, we outline several predictions and questions concerning the effects of larval nutrient deprivation on adult morphology and fitness. In keeping with results on other species, we first expect that adults resulting from a larval nutrient stress treatment would have smaller body mass than those fed an unrestricted diet. Second, we expect that those smaller adults would have a lower reproductive success, based solely on their smaller size. Third, we expect that the allometric relationship between wing length and body mass may change, reflecting either attempts to conserve reproductive potential, or possible changes in wing loading and ability to disperse out of a “bad” or deteriorating environment (e.g., Fric and Konvicka 2002). Additionally, we ask two new questions. First, is there an additional effect of larval nutrient stress treatment on reproduction, beyond that accounted for by body mass? This would occur if stress were accompanied by allometric changes in allocation affecting the relationship between mass and fat reserves or oocyte numbers, for example. Second, is adult survival affected by larval nutrient stress treatment, either due to changes in body mass, or treatment effects independent of body mass? This could occur for example if there were changes in the relative proportion of nutrients allocated to body building and somatic maintenance relative to other functions. These data give us a more complete picture of the within-species response of allocation to variation in the larval and adult food environments, on which to build a theory of allocation trade-offs.
2 Materials and methods
Speyeria mormonia is a univoltine butterfly found in montane-zone meadows in North America. Adults feed primarily on nectar from Compositae, and males and older females feed at mud puddles, dung, and carrion (Boggs and Jackson 1991; Sculley and Boggs 1996; Boggs and Dau 2004). Larvae feed on Viola spp., after diapausing over the winter as unfed first instars.
Larvae used in this study were the offspring of 25 females. Those females were caught in August 1995 from a population near the Rocky Mountain Biological Laboratory, Gothic, Gunnison County, CO, USA. Larvae were transported to Stanford University, where they were maintained in diapause for 4 months, until December 1995. Larvae were then initially started in groups of 36–50 individuals of the same family, placed in net sleeves on potted V. soraria in a greenhouse at 27°C day: 21°C night, with a 15:9 light:dark cycle. As larvae became larger, they were split into smaller groups, down to 5–10 larvae by mid-last larval instar.
Larvae were allowed ad libitum access to V. soraria leaves until the middle of the last larval instar. As individuals reached that development point, larvae from each family were arbitrarily split into two feeding treatments, one of which continued with ad libitum access (“fed” treatment), and the other of which had two medium sized leaves per five larvae per day (“semi-starved” treatment). In practice, larvae in the semi-starved treatment received roughly half the amount of food consumed by larvae in the fed treatment. Given exponential larval growth, a large fraction of the nutrients from larval feeding are obtained in the last instar (Scriber and Slansky 1981).
Larvae spent 4–7 days in the semi-starved vs. fed treatment. Since larvae were reared in groups, it was not possible to track individual development times reliably. Variation in treatment time was due in part to differences in the exact point within the last instar at which larvae entered the treatment, which should be random with respect to treatment. Additionally, it is possible that larvae from the “semi-starved” treatment took longer to pupation than those in the “fed” treatment (e.g., Fischer and Fiedler 2001).
We measured the mass of each adult butterfly to the nearest 0.1 mg in the afternoon of the day of adult eclosion. We measured the length of each forewing to the nearest 0.01 cm; winglength is recorded as the larger of the two wings. Sample sizes for these measurements for females were 67 individuals in the semi-starved treatment (21 families, median family size 3, range 1–8) and 91 individuals in the fed treatment (21 families, median family size 4, range 1–11). Sample sizes for males were 121 individuals in the semi-starved treatment (25 families, median family size 4, range 1–16) and 145 individuals in the fed treatment (24 families, median family size 5, range 3–13).
Adults used in the study of life history parameters were fed ad libitum on a 1:3 v:v honey:water solution twice a day. Males were allowed to mature for several days before being mated. Females were either frozen for dissection or mated on the day of adult emergence. All matings were outcrosses. Both “fed” and “semi-starved” females were mated to “fed” males, to assess the effect on female fitness of female larval treatment. Additionally, “fed” females were mated to both “fed” and “semi-starved” males, to assess the effect of male larval treatment. All females were mated only once, as is generally the case in the field (Boggs 1986). Some males were allowed to mate two times, with 2–5 days between the first and second mating. For all individuals, family number, mate’s family number, and dates of emergence and mating were recorded. 19 semi-starved females from 14 families (median family size 1, range 1–3) were mated and used in the analysis of treatment effects on lifespan and fecundity, along with 34 fed females from 15 families (median family size 2, range 1–4).
Mated females were kept in cages constructed from glass Coleman lantern tops in the manner of Boggs and Ross (1993). Eggs were counted and collected daily. Adult female lifespan was recorded to the nearest day.
We dissected 66 newly emerged females (25 from 15 semi-starved families, median family size 1, range 1–5; and 41 from 19 fed families, median family size 2, range 1–7) and 28 of the mated females upon death, and counted the number of eggs in the ovaries. We calculated the potential fecundity of each female as the sum of the number of eggs laid (if any) and the number of oocytes. Potential fecundity at death in mated females reflects any egg resorption, which occurs under adult nutrient stress (Boggs and Ross 1993). In addition, we qualitatively scored the fat stored in the abdominal cavity on a scale of 0–2; 0=no fat reserves, 1=some fat reserves, and 2=abundant fat reserves. Fat was thus measured relative to body cavity size.
Analysis of variance or covariance using proc GLM in Systat10 was used to evaluate effects of larval feeding treatment on adult morphological and life history traits. Female body mass was natural log transformed for these analyses. While “family” was included in the statistical tests as appropriate, we did not use it to explore heritability or gene × environment interactions, since family sizes were small and “family” includes both common larval rearing environment and genetic components. Rather, we used family to simultaneously control genetics and pre-last instar feeding experience. Allometric relationships are usually defined by a power function (y=axb). In this case we determined the slope, b, for body mass on wing length using the natural log form of the equation (ln y=ln a + b ln x). After testing for homogeneity of residual variances using an F test, we tested for the significant differences between regression slopes, again with an F test.
3.1 Adult morphological traits
Adult morphological traits. See methods for details of “semi-starved” versus “fed” larval treatments
Body mass (mg)
Forewing length (cm)
Body mass (mg)
Forewing length (cm)
Additionally, for both mass and wing length, sexual dimorphism was smaller in semi-starved than in fed individuals. There was a sex × treatment interaction effect when both sexes were considered together using an ANOVA including sex, treatment, family and sex × treatment (ln body mass: F1,392=12.55, P<0.001; wing length: F1,396=5.18, P=0.02).
Allometric relationship between body mass and forewing length by sex and treatment, calculated using ln (body mass)=ln a + b × ln(forewing length)
Within a sex, residual variances were homogeneous (females: F63,87=1.12, n.s.; males: F119,140=1.24, n.s.), allowing tests for differences in slopes of the allometric relationships. For females, individuals semi-starved as last instar larvae had significantly higher slopes than those fed ad libitum, indicating body mass decreased more rapidly with wing length in semi-starved females (Table 2, Fig. 1). Males showed a similar, but non-significant trend (Table 2, Fig. 1).
3.2 Adult female life history traits
Effect of male larval feeding treatment and male mating number on life history traits of their female mates, using an ANOVA with male treatment and mating number
Male larval treatment
Male mating number
Female realized fecundity
Female potential fecundity
Effect on realized fecundity of female traits, using an ANCOVA
ln body mass
Potential fecundity was measured as the sum of realized fecundity and oocytes remaining in the ovaries for mated females, and as oocytes in the ovaries for newly emerged females. Mated females had a significantly higher potential fecundity than did newly emerged females (least squares means ± SE from an ANCOVA with mating status, ln body mass and female larval treatment: mated females: 347 ± 20 (n=28); newly emerged females: 296 ± 14 (n=66); F1,90=4.56, P=0.04). Since this species emerges with a full complement of oocytes (Boggs and Ross 1993), this difference most likely represents detection/ counting error for the smallest oocytes.
Effect of female traits on potential fecundity in newly eclosed adult females, using an ANCOVA (a) with larval treatment and (b) without larval treatment
With larval treatment
ln body mass
Without larval treatment
ln body mass
For virgin females, fat score was greater in those in the fed than in semi-starved treatment (least squares means ± SE from an ANCOVA with treatment and body mass: fed: 1.89 ± 0.06; semi-starved: 1.54 ± 0.09; F1,63=6.94, P=0.01). Additionally, fat score increased with body mass (F1,63=4.15, P=0.04). This indicates that fat storage was greater in larger females independent of larval treatment, even though the fat was scored relative to body cavity size.
4.1 Adult morphological traits
As expected, body mass and wing length were smaller in semi-starved individuals of both sexes. However, female somatic allocation shifted in response to a deteriorating last larval instar food environment, with body mass of semi-starved individuals increasing more steeply with wing length compared to well-fed individuals. This could result in changes in flight capability and metabolism, depending on effects of larval nutrient stress on other aspects of wing morphology (e.g., Kingsolver 1999) and on the relative allocation of mass between abdomen and thorax (e.g., Dudley and Srygley 1994; Dudley 2000; Berwaerts et al. 2002).
Allometric relationships between body mass and wing length in field-caught insects will be influenced by variation in their larval diet, as well as by increased measurement error for body mass due to variation in the exact age at which mass is measured. Thus, we would expect field regression slopes to be intermediate between fed and semi-stressed greenhouse animals. For summer 1995 females in the field, the generation immediately previous to the animals used in the greenhouse experiment, the slope was 1.84 (95% confidence limits: 1.28–2.41; F1,91=42.48, P<0.001) (CLB, unpublished data).
Changes in allometry between female body mass and wing length are consistent with experimental results of Nijhout and Emlen (1998), who showed that allometric relations between various body parts could be altered through resource competition during development. Nonetheless, as for males here, changes in larval diet are not always accompanied by changes in allometry between body mass and wing parameters in other species. For example, Angelo and Slansky (1984) found significant consistent allometric relationships between adult wet body mass and wing area for each of four species of noctuid moths subjected to differing durations of larval starvation. Likewise, Wissinger et al. (2004) found a consistent allometric relationship among body parts within a sex for individuals fed reduced, ambient, or enhanced diets in a caddisfly, Asynarchus nigriculus (Trichoptera: Limnephilidae), although overall body size changed with diet. Unlike S. mormonia, A. nigriculus occur in ephemeral habitats (temporary wetlands), which may influence the consistency under larval food stress of allometric relationships among body parts involved in flight.
4.2 Adult life history traits
Female potential fecundity decreased with decreasing body mass and fat content, but with no effect of larval feeding treatment per se, independent of body mass. In contrast, female adult survival decreased in the semi-starved treatment, but with no effect of body mass. These results stand in sharp contrast to the effect of reducing adult food availability: in that case, dietary treatment affected fecundity, but not survival (Fig. 5) (Boggs and Ross 1993).
Combining the results of larval dietary restriction (the present study) with that of adult dietary restriction (Boggs and Ross 1993), we can draw some conclusions regarding allocation to reproduction and survival in a holometabolous insect with a nectivorous adult stage. With regard to reproduction, these results suggest that the total amount of larval reserves in combination with the amount of nectar available in the adult stage determines reproductive potential. In general, a small adult has the reproductive potential of any other similarly small adult, regardless of why it became small. That is, there appear to be no changes in allocation of larval resources that disproportionately affect reproduction as a result of larval semi-starvation per se, given that we did not observe a larval treatment effect independent of body mass. In contrast, survival is influenced not by body size or adult food intake, but rather by whether the individual was semi-starved during the larval stage. This suggests that there may be something related to the timing of resource acquisition that causes larval semi-starvation to play a major role in adult survival. As one possibility, survival effects could be connected to changes in allometry among adult body parts noted above, and related to timing of developmental pathways involved in production of the adult body.
Taken as a whole, the effects of larval and adult food restriction are reminiscent of a survival/ reproduction trade-off, but one that is dependent on resource allocation patterns across a developmental life-stage boundary. That is, larval food restriction results in the maintenance of reproductive potential at the cost of survival, while adult food restriction has the opposite effect.
The experimental design used here simulates either restriction of food availability or time available for feeding in the larval stage. Other workers have explored effects of larval food quality, usually testing the effects of different larval host plants. For example, Leather et al. (1998) showed in Panolis flammea (Noctuidae) that females reared on a native host plant had higher adult survival than those reared on either of two introduced host plants from different provenances; male survival was unaffected by larval host plant. Consistent with our results, female realized fecundity was determined by lifespan, rather than by larval host plant.
Our results showing decreased adult longevity in the face of larval dietary restriction are, on the surface, contrary to the widespread phenomenon within animals that dietary restriction results in increased longevity (c.f., Houthoofd et al. 2002). However, longevity is generally measured in the life stage that is subject to dietary restriction; our results are across life stages. Additionally, selection experiments for food stress resistance in Drosophila showed that developmental times are increased, allowing accumulation of greater mass from larval feeding and greater lipid stores, which can buffer adult food stress in selected lines (Chippendale et al. 1996; Harshman et al. 1999). Thus, at a minimum, no increase in adult longevity would be expected if larval-derived stores are impacted by larval dietary restriction.
Number of previous matings by a male or male larval dietary treatment had no effect on female fecundity or survival. Such an effect could occur, since females absorb compounds from the spermatophore passed by the male to the female at mating (e.g., Boggs and Gilbert 1979; Boggs 1995). However, the extent to which male-donated nutrients are used by the female should depend on the opportunity for use of such nutrients, including timing of egg maturation relative to the male donation, the female’s adult diet, and the degree to which the spermatophore is absorbed (Boggs 1990). S. mormonia emerge with no eggs mature, and feed on nectar as an adult, giving ample opportunity for use of male-donated nutrients in egg manufacture. Thus, for S. mormonia, compounds donated by males are either not sufficient or not the right compounds to make up for effects of female larval semi-starvation on fecundity and survival. Thus, males do not help buffer females’ fitness against variation in the larval nutritional environment in this species.
Other species show mixed effects of the male larval food environment or mating history on female reproduction, as would be expected if the role of male donations varies among species. For example, in Choristoneura rosaceana (Lepidoptera: Tortricidae), female fecundity was affected by the plant species fed on by their male mate and by whether the mating was the first or second by the male (Delisle and Bouchard 1995). Likewise, female likelihood of re-mating was higher in C. fumiferana for females mated to males with a poor quality diet but fed the same plant species (Delisle and Hardy 1997), although female fecundity and survival were unaffected by male dietary treatment for those females that did not re-mate. In contrast, male body size (presumably affected by both genetics and environment) and previous number of matings had no effect on female fecundity or survival in Zeiraphera canadensis (Lepidoptera: Tortricidae) (Carroll 1994).
The inability of male S. mormonia to buffer the effects on fecundity of female larval semi-starvation, combined with the decrease in sexual dimorphism in body mass, is not consistent with the hypothesis of Leimar et al. (1994) that sexual dimorphism in body mass should increase (with females maintaining mass to a greater extent than males) in the face of low quality larval host plants for monandrous species in which male nutrient donations do not play a major role in female reproduction (see also Gwynne 2004). This hypothesis was supported by tests of changes in sexual dimorphism in body or abdomen mass in both Pieris napi (Pieridae), (a polyandrous species, with a large effect of male nutrient donations on female fitness) (Leimar et al. 1994) and Pararge aegeria (a monandrous species with small male nutrient donations) (Karlsson et al. 1997). S. mormonia is very similar to P. aegeria, both in having a low number of matings by females (~ 1.04 and 1.03 for P. aegeria and S. mormonia, respectively) and a low coefficient of variation in wing length in wild-caught females (~3.3% and 3.8% for P. aegeria and S. mormonia respectively) (Boggs 1986; Boggs 1987; Karlsson et al. 1997). Hence, differences in the extent of normally encountered larval host variability should not explain the discrepancy between species. As noted above, however, it is not clear whether females are conserving abdomen mass relative to thorax mass under larval nutrient stress, given the maintenance of potential fecundity, but reduction in fat score. Either such a change in allometry between thorax and abdomen, or the fact that the experiments of Leimar et al. (1994) and Karlsson et al. (1997) stressed larvae for the whole lifespan could account for the discrepancy in results. Nonetheless, the reasons for apparent differences among these studies in conservation of reproductive potential by females relative to males under dietary stress remain to be explored.
We thus see a suite of life history and morphological trait changes in response to nutrient stress in different life stages of a holometabolous insect. Carbon is abundantly available from the adult nectar diet, and determines fecundity through its immediate use in the manufacture of eggs. At the same time, larval resources are necessary for growth and development of a normal sized and proportioned adult with adequate reserves of compounds not available in the adult diet. Resultant body mass impacts fecundity. However, survival is determined by some other aspect of larval feeding treatment, presently unknown, but likely due to allocation changes in response to stress. This type of understanding of the effect of stress at different life stages will be critical to an integrated understanding of the ability of organisms to buffer fitness, hence population size, in increasingly variable environments.
We thank Craig Fee and David Stiles for help rearing and dissecting butterflies. Ilkka Hanski, Bengt Karlsson, Craig Osenberg, Ward Watt, and the reviewers commented helpfully on the manuscript.